Sign In

Communications of the ACM

ACM TechNews

IBM Has a Watson Dilemma

IBM Watson logo and heart rate graph, illustration

IBM's Watson artificial intelligence (AI) system is falling short of some expectations for its outcomes in the fight against cancer, with many users citing inaccuracy, the AI's susceptibility to error due to a lack of data, and the evolution of cancer therapies outpacing the ability of human trainers to upgrade Watson. IBM says Watson offers valuable cancer-care benefits, such as helping physicians keep up with medical knowledge.

For an AI to recommend fine-tuned personalized cancer treatments, the software requires training with data on past treatments, including patients' medical histories and treatment outcomes, which is often recorded in differing format and may be incomplete or inconsistent. IBM's planned integration of medical-claims databases and electronic health records has not yet been realized, partly due to the task's complexity.

IBM's John Kelly remains positive on Watson's potential, saying the system's treatment recommendations have changed the course of care for as many as 8,400 patients thus far.

From The Wall Street Journal
View Full Article – May Require Paid Subscription

Abstracts Copyright © 2018 Information Inc., Bethesda, Maryland, USA


No entries found